______o_c_EA_N_O __ LO_G __ IC_A_A_C_T_A __ 19_8_8_-_v_o_L_._1_1_-_N_o_2~~~-----

Current meler array Dynamics Bottom nepheloid layer Deep layer variability Variability Porcupine abyssal plain in the Eastern North Atlantic: Mouillages de courantomètres Dynamique Couche néphéloïde profonde the EDYLOC experiment Variabilité Plaine abyssale Porcupine

Annick V ANGRIESHEIM Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER), Centre de Brest, B. P. 70, 29263 Plouzané, France.

Received 7/4/87, in revised form 19/11/87, accepted 24/11/87.

ABSTRACT The ÉDYLOC ("Étude DYnamique LOCale") experiment was carried out by IFREMER-Centre de Brest in 1981-1982 in order to describe the dynamics and evaluate the space and time scales of current variability in the deep layer of the Porcupine abyssal plain (North East Atlantic). Six moorings separated by 11 to 73 km were deployed in a flat bottom area near 47°N, 14°30'W. The water depth was 4780 m. The results are described here. A hydrological survey, with nephelometry profiles, shows that a nepheloid layer exists at each station with the strongest signal associated with the only observed bottom mixed layer in the north-western part of the array; it is also there that the currents are the strongest throughout the measurement period. Over the 11 months of record (data return: 96.2%), the horizontally averaged mean speed is 1.26 cmjs at 4000 m and 1.50 cm/s at 10 m above the bottom with a predominancy of the westward-component. The correlation between the two levels is very high, which means that the deepest level dynamics is not influenced by the bottom effect to the extent that one might expect in a bottom boundary layer. A bottom intensification is observed on KM (kinetic energy of the mean flow) and on KE (eddy kinetic energy) and, for both levels, the currents appear more energetic than for other previous experiments. The zero-crossing of the auto-correlation functions is 20 days for the U-component at both levels and 66 and 33 days for the V-component at 4000 and 10 m above the bottom respectively. At both levels, the zero-crossing of the transverse velocity correlation function is 27 km. Oceanol. Acta, 1988, tt, 2, 149-158.

RÉSUMÉ Variabilité dans la couche profonde de l'Atlantique Nord-Est : l'expérience ÉDYLOC L'expérience ÉDYLOC (Étude DYnamique LOCale) a eu lieu en 1981-1982 dans le but de décrire la dynamique et évaluer les échelles spatiales et temporelles de la variabilité du courant dans la couche profonde de la plaine abyssale Porcupine dans l'Atlantique Nord-Est. Six mouillages séparés de 11 à 73 km ont été installés dans une région à fond plat autour de 47°N, 14°30'W sur une profondeur d'eau de 4 780 m. Les résultats en sont décrits ici. Une reconnaissance hydrologique, avec des profils de néphélométrie, a montré l'exis­ tence d'une couche néphéloïde à chaque station, la plus intense étant associée à l'unique couche mélangée de fond observée dans la partie nord-ouest de la zone; c'est également à cet endroit que les courants ont été les plus forts pendant toute la période de mesures. Sur les onze mois de mesures (retour des données : 96,2 %), la vitesse moyenne, moyennée horizontalement, est de 1,26 cm/s à 4 000 m et 1,50 cm/s près du fond, avec une prédominance pour la composante Est-Ouest. La corrélation entre les deux niveaux est très élevée, ce qui indique que l'effet de couche limite n'est pas si important.

0399-1784/88/02 149 1 0/$ 3.00/ (Cl Gauthier-Villars 149 A. VANGRIESHEIM

Une intensification est observée sur KM, énergie cinétique moyenne et sur KE, énergie cinétique turbulente, et aux deux niveaux, les courants sont plus énergétiques que lors d'expériences précédentes. Le premier passage à zéro de la fonction d'autocorrélation moyenne est à 20 jours aux deux niveaux pour la composante U et à 66 et 33 jours pour la comp-osante V à 4000 rn et au fond, respectivement. Aux deux niveaux, le premier passage à zéro de la fonction de corrélation transverse est 27 km. Oceanol. Acta, 1988, tl, 2, 149-158.

INTRODUCTION Another approach to the characterization of the influ­ ence of the bottom in the deep layer is to examine The EDYLOC Experiment ("Étude Dynamique its hydrological fine structure and the existence of a Locale"), was conducted in order to evaluate spatial nepheloid layer. As a result of homogenization due to and temporal scales of deep current variability in the turbulence above the bottom, a bottom mixed layer eastern North Atlantic Porcupine abyssal plain. Long­ (BML) appears in the lower part of the water column, lerm current measurements were carried out with the its thickness depending on the strength of this dynamic deployment of six moorings, with two levels of mea­ activity. It cannot be directly related to the theoretical surements each (4000 and 10 rn above the bottom); a Ekman layer because a BML may be the result of CTD survey was performed during the moorings reco­ superposition of local old and recent BMLs and very cruise. The two level results will allow a compari­ advected BMLs. Armi and Millard (1976) studied the son of the ocean interior dynamics with that of the BMLs over the Hatteras abyssal plain; they found that benthic boundary layer. the BML thickness was correlated with the daily mean speed of the current and its vertical extension was Briefly, the characteristics of the benthic boundary around 6 times the theoretical Ekman layer thickness. layer will be reviewed. Theoretically, in close proximity Nevertheless, the existence of a BML is the signature to the bottom, a "logarithmic layer" must be found, of a mixing activity on the bottom which makes this where the current speed is assumed to decrease as a layer dynamically different from the upper layer. logarithmic function to reach zero on the bottom; this theory supposes that shear is constant in the layer and The bottom mixed layer is usually associated with an intense particle resuspension whose vertical extent may that the Coriolis forces may be neglected compared with viscosity. Above this layer, the Coriolis forces sometimes be much larger than that of the BML itself. cannot be neglected and the Ekman theory is relevant; This layer above the bottom, where the suspended particle concentration is higher due to the presence of the current describes an "Ekman spiral" with its cha­ racteristic veering. In the oceans, ali the Ekman layers the bottom, is the "bottom nepheloid layer" (BNL). are turbulent. The logarithmic layer is the one-tenth As for the BML, it may result from the superposition deepest part of the Ekman layer. In deep oceans, the of locally resuspended particles and advected particles. The different aspects of nepheloid layers are reviewed order of the Ekman layer thickness is 10 rn and that by McCave (1986). of the logarithmic layer is 1 rn (for a velocity of 4 cmjs and a friction velocity u* of 0.2 cm/s). The theoretical It was with the aim of describing this environment that difficulty is to link these two layers mathematically. a CTD survey (with nephelometry) was planned during It is more difficult to observe these layers in the oceans, the moorings recovery cruise. particularly in deep oceans. In fact, the logarithmic The moorings were launched during the profile has been observed, but the existence of an "EDYLOC 81" cruise, in May 1981, aboard the R/V Ekman layer with its associated veering bas been less Le Noroît and recovered in April 1982 during weil established (Bowden, 1978). This is probably due . "EDYLOC 82" aboard the same ship. to the non-stationary currents (tidal oscillations) To exclude local topographie effects, the region selected encountered in the areas of observations. In the loga­ was a deep abyssal area (4780 rn) over the Porcupine rithmic layer, the time scale (of the order of zfu* which abyssal plain around 47°N, 14°30W, where the Tourbil­ is around 10 minutes) is small compared with the tidal lon experiment took place in 1979 (Fig. 1). period which allows the establishment of the loga­ The moorings array was designed as a cross, the rithmic profile. The time scale of the Ekman layer is branches of which were respectively parallel and normal of the order of the inertial period which is comparable to the continental margin. To obtain a maximum of to the tidal periods. Then, the tidal or inertial oscilla­ mooring pairs to compute the spatial correlations, the tions disturb the establishment of the Ekman layer. moorings were separated with increasing distances, the In our experiment, the deepest level (10 rn off the smallest being 11 km and the largest 73 km. A similar bottom) is quite surely above the logarithmic layer but experiment was conducted elsewhere during the same could be in the upper level of the Ekman layer if it period by I.O.S. (Institute of Oceanographie Science, exists. Comparison with the 4 000 rn results will allow UK), the results of which (Saunders, 1983) will be us to determine whether this is the case. compared with ours.

150 DEEP LAYER VARIABILITY IN THE EASTERN NORTH ATLANTIC

Each mooring carried two Aanderaa current-meters, This deviee (Fig. 2) has been developed at IFREMER the deeper at 10 rn off the bottom and the shallower by Gouillou as a sensor dependent on the CTD both at 4000 rn depth, i. e. around 800 m off the bottom. for its power supply and its data acquisition system; This level was chosen as representative of the ocean its measurements are recorded simultaneously to those interior and will be used in a comparison with the near­ of temperature, salinity and oxygen. It is an adaptation, bottom level; furthermore, this depth is a common for deep pressures, of a shallow-water instrument built level with previous experiments in this area: Neads, by Prieur at the Laboratoire de Physique et Chimie Tourbillon 79 ... Marines of Villefranche-sur-Mer (France). It measures During the recovery cruise, CTD profiles were carried the optical scattering, by suspended matter, of a bearn out in order to describe the hydrological environment. near an angle of 4°. The ratio of scattered light over The Neil-Brown CTD was equipped with a nephelome­ direct light is linearly related to the optical scattering ter, a prototype built at IFREMER-Centre de Brest coefficient and to the particle concentration. The cali­ whose first tests on this cruise were satisfactory. bration versus diffusion coefficient (metres-1) has been obtained by comparison with the Prieur's instrument (Vangriesheim, Gouillou, 1987). To obtain the relation­ ship with particle concentrations, a delicate in situ calibration is needed with dry weight. Because of the low concentrations encountered in the deep ocean, this relation is not yet obtained. Thus, the nephelometry data will be presented here as nondimensional values which are those of the scattered lightfdirect light ratio.

RESULTS

Hydrological environment The hydrological structure can be studied from the 1 1 1 1 CTD stations carried out around the moorings. 3 Figure 3 shows, as an example, profiles of potential 47°"" 2* ~,J,.., '\/~;w temperature, salinity, and oxygen at station 19 which reached the bottom. This figure enables the different /1x y.> water masses to be identified: ,~.., r- !~'/ '\ - '\.~/ - a surface layer affected by seasonal fluctuations of ,'b/ '). / sorne tens of metres; / 4 - Northeast Atlantic central water down to around r- }If, - ~~~/ 5 500m; r>:>~/ '/ / - Mediterranean water characterized by a salinity *'6 f-47°nn maximum and small temperature inversions around 1000 rn; 15°oo 14°30 14"oo 1 1 1 1 - Labrador Sea water around 1700 rn with its mini­ mum of salinity; Figure 1 Location and schema of the moorings array of the EDYLOC 81-82 experiment.

Receiving lens Transmitting lens

Electronic container Light source container Windows

0 10 50 cm Figure 2 The nephelometer built at IFREMER which was instal­ Scala led on the CTD.

151 A. VANGRIESHEIM

4.0 5.0 6.0 7.0 02 potential temperature· and salinity decrease slowly 35.00 35.50 SALINITY down to the bottom. This absence of bottom mixed layer linked with a weaker signal of nephelometry at 0 3.0 6.0 POT. 1EMP.(oc) 0+-~~~~~_L~~~~~~ these stations could indicate that the boundary effect on the dynamics is weaker than at station 16 where the effect of the bottom is well-marked. After these observations, it is evident that the upper­ 1000 most current-meter ( 4000 rn) is above the nepheloid layer which is, itself, thicker than the slight and thin bottom mixed layer observed at only one station. The deepest current-meter (10 rn off the bottom) is in the 2000 nepheloid layer and in the BML when it exists. We may thus expect sorne differences between the dynamics of the two levels.

• CTD Stations '5...... 16 ,:4*- 1-19 3000 1- 47"30 Currentmeter moorings 5 ·~r * ~-· \. ~7 13. g* MILES ... t ~ 1"10 :0 0 20 • 7 "0 L...... J..__ ~8 2 "' 4Too ~ 20Km :; 4000 ~, 0: SCATTERING 0 .015 .030 15•oo 14"oo 15 16 14 19 17 18 i ~~==~~~==~0 1 ç--~{ 5000 Figure 3 1000 Vertical profiles of potential temperature, salinity and oxygen concen­ 1 f tration at station 19 (47°34.43' N, 14°07.52' W). 2000

:0 - a layer with a slight deep salinity maximum around :s 3000 w 2300 rn which is usually attributed to Norwegian Sea 0: water influence but which could be due to that of :::> 4000 Mediterranean water according to Harvey and Glynn lli0: (1985); (l. Figure 4 \ 5000 - the Northeast Atlantic bottom water of Antarctic origin where salinity and temperature decrease slowly Vertical profiles ofnephelometry at stations 15, 16, 14, 19, 17, 18. to the bottom. .005 .010 SCATTERING During the CTD survey, only 6 stations with nephe­ lometry measurements were carried out down to the 34.85 34.90 SALINITY bottom (the other CTD casts stopped at 4000 rn). 2.0 2.5 POT. TEMP. (•c) The vertical profiles for this parameter are drawn on 2500+-~~~~~~~~~~-7~ Figure 4; the upper part has been truncated because of the very high values at the surface. As the nephe­ lometer has been developed in order to measure low particle concentrations in the deep ocean, its response 3000 in the high concentrations of the euphotic layer is non­ linear; consequently, the high surface values should not be compared to the remaining profile. The strong peaks which sometimes appear are due to large particles 3500 which saturate the scattering signal. Besides the upper layers, the scattering signal is rather homogeneous, except near the bottom where an 4000 increase of the signal is observed. This reveals the presence of a nepheloid layer, the top of which lies between 4100 and 4400 dbar (thickness of 750 to :0 500 rn) except at station 16 where the layer top is at :s w 4500 4700 dbar and where a strong signal is observed close 0: to the bottom; at this station, the nepheloid layer is ~ then the thinnest (200 rn) but the strongest and it is 8: the only station where a bottom mixed layer (around 5000 90 rn thick) appears on the potential temperature and Figure 5 salinity profiles (see expanded deep profiles on Fig. 5). Expanded deep vertical profiles ofnephelometry, potential temperature At the other stations, no bottom mixed layers appear; and salinity at station 16 (47°31.50' N, 14°45.80' W).

152 DEEP LAYER VARIABILITY IN THE EASTERN NORTH ATLANTIC

Currents correlation between high currents and existence of BML and BNL is strong at the end of these measure­ The data set ments, it may be supposed that this deep structure lasted throughout the period where M2 currents were Current and temperature data were collected with Aan­ high, i. e. throughout the 7 months of measurements. deraa RCM5 current-meters. Rotor, compass and tem­ perature sensors were calibrated at IFREMER before Statistics deployment. Ali instruments were equipped with an From the unfiltered data, the maximum speed encoun­ expanded temperature scale whose resolution is 8 m°C. tered lies between 12.4 and 14.4 cmfs at 4000 rn and Currentmetres, at 4000 rn, had a pressure sensor to between 14.4 and 16.3 cmfs at 10 rn off the bottom, measure the mooring line excursions which were, in thus indicating higher maximum speeds near the bot­ fact, insignificant. tom. For U-components, values at 4000 rn vary from The data return was very good. Measurements lasted -12.9 to 12.7 cmfs and from -15 to 14.6 cmfs at from 19, 20, or 21 May 1981 to 6, 7 or 8 April 1982. 10 rn off the bottom; for V-components, the values lie Only 2 of the 12 data series have short gaps: at mooring M3, speed data at 4000 rn starts only on 19 July, and at mooring 6, temperature data at 4000 rn stop on 9 January. Thus, the data return is 98.5% for speed data and 97.7% for temperature datai. e. 96.2% to have the two parameters simultaneously. The data sample interval was 1 hour. The data have been processed and presented either unfiltered or filtered, depending on the calculation con­ cerned. To remove high frequency oscillations, such as tide and inertial oscillations, a Lanczos filter with a eut-off period of 2 days has been used. EDYLOC82 General description of the circulation 4000m A data report (Vangriesheim, 1986) presents ali the results by means of graphies such as speed, east and north components (U and V), temperature versus time, stick-plots, progressive vector diagrams, energy spectra and different statistical values. It may be consulted for complementary details. To give an insight of the pattern of the circulation deduced from our measurements, Figure 6 shows the PVDs at 4000 rn and 10 rn off the bottom on maps where the starting point of each PVD coïncides with the location of the mooring concerned. Except at mooring 3 where measurements at 4000 rn start two months Iater than the others, these fictitious trajectories show a strong correlation between the two levels. At EDYLOC 82 moorings 1, 2, 3, which are the nearest, the westward 10 m off bottom component is predominant throughout the duration of the measurements. At M4, the northward component is always predominant. For moorings 5 and 6, no pre­ Figure 6 dominance appears. Progressive vector diagrarns obtained for the 6 rnoorings at 4000 rn The stick-plots of Figure 7 show the time-fluctuations and at 10 rn above the bottorn. of the current at moorings M2 and M5, where it was respectively the strongest and the weakest. For these Mz 4000 m two extreme cases, the correlation between the two 10oo[N;;> rr, w:lllv --7rl47 -~"'. ''»lh1' ..... '7~ j levels is evident. Evident also but not so strong is the g -100 s

153 A. VANGRIESHEIM

between -14.1 and 12.5 crn/s at 4000 m and between M2 M3 TI.v Ccmts) KM, KE(cm2ts2) 4.8 -12.8 and 14.1 cmjs at 10 m off the bottom. Thus, 4000m 4000m 5.9 3.7 the two components seem to reach maximum values of M1 4.5 1.9 the same order. . ·· 5.4 M4 Ms Ms 0,9 Figure 8 presents mean velocities, for both levels, at 3.5 -0.62 0.2 the six points of measurements. As in Figure 6, the o. 33r--,,..,..,...,.,....,.,-:-,-,;=-l 5.5 r:-::::-:-:-:-:-:::-:-:-::=-1 Ma MEANVAWES Ms MEANVAWES westward-component appears to be predominant at o.1o COJ -1.25 (0.34) 0.8 (KM) 0.8 both levels for Ml, M2, M3, as does the northward­ -1.10 (V) 0.13 (0.47) 3.5 (KE) 5.9 (0.7) component at M4. The slight veerings (1 o to 10°) U,V (cmts) M2·3.29 MJ KM, KE (cm2/s2) 0.37 -3 12 between the two levels seem to have no significance in bottom Ml -o:os bottom terms of Ekman veering: when the two vector directions 2.6 -2.2 9.2 are distinct, the veering is twice on the right and twice 0.20 M4 M4 M5 -0.12 M5 1.8 on the left (at M2 and M3 where the currents are the 1.90 6.0 -0,76 0.4 highest). 6 3 Ms o.so,.,.M"""EA::-:-:-N.,.,.V...,'A.,...W"'E=s~ Ms ' MEAN VAWES 0.60 -1.49 (0.381 0.7 1.1 -I.OS 0.19 (0.48 6.4 8.8 (0.9)

N Figure 9 V, V and KM, KE for each mooring at 4000 m and 10 m above the bottom and mean values (i.e. averaged over the 6 moorings) with errors. i •1 calculation has been made also within frequency bands; it shows that no higher polarization appears at any Ms frequency band and that ali the energy is concentrated in the low frequency range with almost the same pola­ cm/s Ma 0 1 2 3 rization as in the mean signal (Mercier, pers. comm.).

\ ' 4000m Bottom Frequency analysis As regards to the energy content of the data, Figure 10 Figure 8 shows, as an example, the kinetic energy spectrum The 11-month mean velocity field at 4000 m and 10 m above the obtained from unfiltered data at both levels for Ml. bottom. Strong signais can be identified at semi-diurnal and diurnal frequencies and at inertial frequency (inertial period in this area is around 16h30). At both levels, The lowest mean velocities are O. 7 and 0.9 cm/s the energy associated with semi-diurnal tide is higher (4000 m and bottom respectively) at MS and the highest are reached at M2 with 3.1 and 3.3 crn/s M1 400Qm (4000 m and bottom respectively) which shows, as for PERIOD h- maximum speeds, higher values near the bottom. Figure 9 summarizes, on a grid resembling the mooring array, the mean values for U- and V-components, KM and KE for each filtered data serie as weil as their averages (with errors). An estimation of the error has been obtained based on U' 2, V' 2, the time scales found for U and V at each leve! (see section concerned with autocorrelations) and considering that, as the hori­ zontal correlation length is 27 km, the average is made 0.001 0.1 with 3 independent data sets: Ml-M2-M3-M4, MS FREQUENCY Cycles/h and M6. The method used for this estimation is taken from Bendat and Piersol (1971). PERIOD h M.J 10 m above bottom 10 1000 At both levels, the error is higher for the V-component ~1,.c due to the fact that the integral time scale is longer for § ~ 103 this component. 102 z~ From Figure 9, it may be deduced that the horizontal 1:!10, average speed is 1.26 crn/s at 4000 m and 1.50 cmfs b near the bottom. ~ 10° I ao .. The values of U' 2, V' 2 and U' * V' have been used ~ 10·1 TOTI\L ~ SPECTRUM to determine the principal axes of the fluctuations. At 52 4000 m, the principal axis is oriented along 003° i.e. 0.001 0.1 almost northward, and 57% of the energy is in this FREQUENCY Cycles/h rn direction. At 10 above the bottom, the direction is Figure 10 018°, i.e. slightly most eastward and the energy is not Total, clockwise and anti-clockwise spectra obtained from unfiltered more polarized in this direction (57% also). Such a data at mooring 1 (4000 m and 10 m above the bottom).

154 DEEP LAYER VARIABILITY IN THE EASTERN NORTH ATLANTIC

Table 1 Confidence levels (at 90%) and numbers of degrees of freedom for the energy spectra of Figure 11.

4000m 10 rn above bottom Period band limits limits days f•KE (cmfs) 2 N f•KE (cm/s) 2 N 626 to 208 2.43 1.15 to 8.91 6 3.08 1.47 to 11.31 6 208 to 125 4.07 1.94 to 14.9 6 5.24 2.50 to 19.23 6 125 to 69 2.64 1.51 to 6.08 12 3.53 2.01 to 8.10 12 69 to 48 2.54 1.45 to 5.84 12 3.61 2.06 to 8.29 12 48 to 30 1.63 1.08 to 2.83 24 2.39 1.57 to 4.14 24 30 to 19 0.81 0.57 to 1.25 36 1.66 1.17 to 2.57 36 19 to 12 0.68 0.51 to 0.94 60 1.37 1.04to 1.91 60 12 to 6 0.55 0.46 to 0.67 156 1.51 1.27 to 1.84 156 6 to 2 0.43 0.40 to 0.48 576 1.39 1.27 to 1.54 576

Table 2 EDYLOC 81-82 EXPERIMENT First zero crossing, integral time scale, square integral time scale from ,---, averaged autocorrelations for U, V, T. 1 1 5 1 1

N 1 1 --At 4000rn First zero Integral time Square integral (,) Q) : 1 -----At JOrn above bottorn crossing sc ale time scale "' 1 : (days) (da ys) (days) N' 1 1 E4 4000m 20 9 8 u u Bottom 19 8 5 _.J'-j ~ 1 4000m 66 17 14 U) 1 v Bottom 33 12 10 z 1 r----- 1 ~3 1 1 4000m 47 14 19 l__ ï T 1 Bottom 23 9 6 >- 1 1 1.9 1 1 0:: 1 w 1 1 EDYLOC 81- 82 EXPERIMENT z 1 W2 1 1l__ __ 1.0 AUTOCORRELATJON FOR U- COMPONENT 1 z (.)>­ z I' __ J ·----. 0 w ~__j Ôl w w 0:: 0:: o::0.5 -- At4000m Li.. 0 ----- At JOrn above bottorn ug ::::> 500 100 50 10 2 <{ PERIOD ( Days) o~~--~---r--~~~~r--.---.--- Figure 11 1.0 Average kinetic energy at 4000 m and 10 m above the bottom in energy z preserving form (obtained from filtered data). 0 AUTOCORRELATJON FOR V- COMPONE NT ~ than that of inertial oscillations. For ail moorings, tidal __j w energy at 4000 m is lower than near the bottom but no 0:: 0::0.5 -- At4000m such significant tendency appears for inertial oscillation 8 ----- At JO m above bottom energy. g ::::> To examine low-frequency fluctuations, horizontaily <{ averaged energy spectra have been calculated at each level from filtered data. The spectra at both level are 10 20 presented in Figure 11 in variance preserving form. Frequency averaging has been made over severa! adja­ Figure 12 cent frequency bands to obtain a better estimation of Autocorrelation functions for U- and V-components at 4000 m and 10 m above the bottom. the result. Table 1 shows the limits of these averaged bands as weil as their f * KE content, the confidence or sorne seasonal signal and is now a rather usual levels (at 90%) and the number of degrees of freedom feature. in each period band. It may be noticed that the energy is higher for the Time and space scales bottom at ali frequencies; this intensification is greater Averaged autocorrelation functions have been calcu­ for periods less than 30 days where the bottom energy lated for U- and V-components at both levels. In exceeds by a factor of two the 4000 m level energy. Figure 12, which presents the results, it appears that This is statistically significant because, as shown in the time scale for the V-component is much longer than Table 1, there is no overlap between the confidence for the U-component. This is shown more precisely on limits of the two levels due to the high number of the Table 2, which summarizes the values of the first degrees of freedom at these periods. zero crossing, the integral time scale and the square A broad maximum appears at both levels between 208 integral time scale for U, V and T at each level. The and 48 days, particularly between 208 and 125 days. autocorrelations have been calculated with a maximum This peak is in the period band affected by the eddies time lag of 150 days. The integral time scales were

155 A. VANGRIESHEIM evaluated over the interval bounded by the first zero EDYLOC 81 • 82 EXPERIMENT crossing, and the square integral time scale was calcu­ 1.0 TRANSVERSE CORRELATION lated up to 150 days. From this table, it may also be z: noticed that for the 3 parameters, the time scales are 0 -At4000m· ~ ...... At IOm obove bottom less at 10 rn off the bottom than at 4000 m. This is the ..J w same feature as that previously observed on the spectra o:0.5 0: where the bottom intensification was higher for short 0 periods. (.) At the opposite, at Tourbillon (Mercier, Colin de Verdière, 1985), the time scales increase with the depth down to 3000-4000 rn (the two levels were aver­ aged in the calculations), where the square integral time 1.0 LONGITUDINAL CORRELATION scale is 11 days for boih U- and V-components which ' z: ' ' is an intermediate value between those found for U 0 ' '•, --- At 4000 m and V at EDYLOC. ~ ' .., ~----- At 10 m obove bottom ..J ' The horizontal scales of the low-frequency currents ~0.5 \ 0: \ have been obtained by computation of the transverse 0 ' \ (.) \.-,.,-•..... , and longitudinal correlations for ali the pairs of data ·-- series at each level. Then, the results have been ave­ raged in each band of 5 km; the results are shown in oL------,25------5~0------~75~­ Figure 13. The transverse correlation bad its first zero SEPARATION (Km) crossing at 27 km at both levels and, as is usually Figure 13 found, the longitudinal correlation never reaches zero. Transverse and longitudinal correlation functions calculated at 4 000 m This horizontal scale lies between those found at and 10 m above the bottom. 3000 rn for Tourbillon (32 km) and on Madeira abyssal plain (Saunders, 1983) at 10 rn above bottom (22 km). currents were the strongest (station 16); at this station, this observation might support the idea of a local process of generation of the BML and BNL. At the DISCUSSION AND CONCLUSIONS other stations of the area, the thicker observed nepheloid layer is more likely a stack of discrete BNLs Hydrological environment and suspended particle generated elsewhere and advected along density surface. According to one reviewer of this paper, this generation The CTD survey performed around the moorings array mode would prevail also for the station 16, considering revealed that, despite a very flat bottom topography, a that the observed nepheloid layer thickness (200 rn) is nepheloid layer appears on ali the vertical profiles. too high compared with the average thickness (50 rn) Nepheloid layers have been observed for a long time of locally generated BNLs deduced by bottom Rn-222 in the West Atlantic (Eittreim, Ewing, 1972) and in the whole Atlantic (Biscaye, Eittreim, 1977), where they profiles (Gurbutt, Dickson, 1986). In this case, the are closely linked to the Antarctic bottom water flow. origin of the BNL observed at station 16 must be due to a topography outside the array at a depth In the Northeast Atlantic, they were observed more corresponding to that of the maximum light scattering recently, particularly over rough topographies or slopes measured at this station. More data would be needed (Vangriesheim, 1985, Fig. 4; Dickson, McCave, 1986; Nyffeler, Godet, 1986) but they bad never been observ­ to confirm this assertion. ed over the Porcupine abyssal plain. In these cases, Mean circulation they are not directly linked with a water mass flow but are caused by dynamical processes over the bottom. Concerning the current measurement results, it appears The feature found here over the plain is rather different that the mean current is more westward than usually from that observed over a slope. On the EDYLOC f ound in this area. profiles, the nepheloid layer is marked by a smooth As for the current measurement results at 4000 rn, the increase of the signal down to the bottom; it is only in westward component is higher than in most of the the last tens of metres that a stronger signal appears. previous experiments conducted in the Eastern North On the slope (Vangriesheim, 1985, Fig. 4), it is charac­ Atlantic (Dickson et al., 1985); this is not the case terized by a more homogeneous layer separated from for the northward component which is Jess in our the upper one by a strong gradient and which coïncides experiment. The resultant at 4000 rn at EDYLOC is with a well-marked bottom mixed layer. This set of towards west-north-west, whereas it was north-east at data is too small to understand the behaviour of the Tourbillon in the same area in 1979-1980 (Mercier, bottom nepheloid layers over EDYLOC area; it is yet Colin de Verdière, 1985). This discrepancy between two enough to know that the uppermost current meter of consecutive experiments made in the same place shows the moorings is above the nepheloid layer. again that a one year time-average of the current is The important point to be noted is the fact that the not enough to be statistically significant. Nevertheless, area where the only BML bas been observed linked the existence of a northward component is consistent with the strongest nepheloid layer is that where the with the Antarctic origin of the bottom water. The only

156 DEEP LAYER VARIABILITY IN THE EASTERN NORTH ATLANTIC southward weaker mean current observed at mooring 6 Nevertheless, KE values at 4000 m are higher at could be explained by the influence of topography to EDYLOC than in previous experiments in the same the south-west of this mooring. basin and, in addition, higher than those obtained at The near-bottom values may be compared also to the same place during Tourbillon 79-80; as for KM, other results: in most experiments in the same area, the EDYLOC values at 4000 m are also the highest the U-component is not so intense at 10 or 50 m above except for the mooring 0 (MAFF) data. the bottom as in the present case. Concerning the near-bottom results, EDYLOC KE values are of the same order as those obtained at the Mean flow kinetic energy and eddy kinetic energy same place during Tourbillon but are higher than those obtained elsewhere in the same basin. In order to compare the values of KE and KM with those of other experiments, Tables 3 and 4 summarize Bottom intensification sorne previous results obtained in almost the same conditions i. e. over an abyssal plain with at least the Concerning kinetic energies, it may be seen on Figure 9 same depth, excluding experiments conducted further that KM is higher on the bottom than at 4000 m for north than EDYLOC or conducted over topography. ali the moorings but M6, and that bottom KE is higher From these tables, the EDYLOC values are in agree­ for the six moorings. On averaged values, it is clear ment with the energy decrease toward the south already that this intensification is statistically significant for noticed by Dickson et al. ( 1985). KE; concerning KM, though it is not statistically

Table 3 Comparison of 4000 m KE and KM ED Y LOC values with other experiments carried out over at /east the same depth in the North-East Atlantic.

Water depth KE KM Experirnent (m) cm2js2 crnz;sz Source EDYLOC 4780 5.9 0.8 Same area Tourbillon 4800 3.3 0.3 Colin de Verdière and Mercier (IFREMER) Same basin Mooring 0 4820 5.62 2.61 Dickson 49 10.4N 1544.6W (MAFF*) Neads 7 4995 2.1 0.12 Maillard 47 OON lOOOW (IFREMER) Neads 5 4760 1.02 0.50 Dickson 46 06N 1709W (MAFF*) • MAFF: Ministry of Agriculture, Fisheries and Food, Fisheries Laboratory, Lowestoft, Suffolk, UK.

Table 4 Comparison of near-bottom KE and KM EDYLOC values with other experiments carried out over at /east the same depth in the North-East Atlantic.

Water depth KE KM Ex periment (m) cm2js2 cm2js2 Source EDYLOC 4780 8.8 1.1 Same area Tourbillon MlO 4775 Vangriesheirn 20 m off bottorn 9.6 0.3 (IFREMER) 50 m off bottorn 9.6 0.3 Same basin Mooring 0 4820 5.63 2.41 Dickson 50 m off bottorn (MAFF) 49 10.4N 1544.6W Madeira abyssal plain 33N 22W 5300 3.6 0.01 Saunders 10 rn off bottorn (lOS**) 81-14 50 rn off bottorn 4912 2.83 1.12 Dickson 81-16 50 m off bottom 5296 2.05 1.78 (MAFF) 81-17 50 rn off bottorn 5027 2.38 0.04 Canary abyssal plain CYl 5200 1.76 0.46 Vangriesheim 10 m off bottorn (IFREMER) 24 49.6N 2502.9W CV2 4885 1.17 1.12 10 m off bottorn 19 14N 2947.7W •• lOS: lnstitute of Oceanographie Sciences, Wormley, Godalrning, Surrey, UK.

157 A. VANGRIESHEIM significant, this increase from 4000 m to the bottom Time and space scales for almost ali the moorings is worth pointing out. As for the time scales, they are consistent with other A bottom intensification has been already observed in North-East Atlantic results but an interesting feature other places and at Tourbillon between 3000 and is the difference between U and V time scales (Tab. 2). 4000 m (Mercier, Colin de Verdière, 1985). However, The V time scale is much longer than the U one which there is no evidence that the one observed here is due was not the case for the Tourbillon experiment. No to the same processes. It would be interesting to know satisfying explanation can be given. up to which level the downward intensification This set of data allowed the evaluation of the horizontal observed at EDYLOC and at Tourbillon could be scales in the deep layer of the Porcupine abyssal plain, found; better vertical resolution is needed between a task not done before employing current measure­ 3000 m and the bottom to determine this. Besides, here, ments. The horizontal scale (27 km) is found to be the this intensification seems to be slightly higher for the same at both levels. The deep spatial scale appears to V-component than for the U-component. be the same as that of upper levels or that of the The average ratios KE/KM are almost identical near southern area. An attempt was made to find assyme­ the bottom (7.8) and at 4000 m (7.4), which means tries in the spatial correlation. To this end, horizontal that KE and KM seem to be intensified in the same correlations were calculated separately with the set of proportion. The intensification of KE, which is the moorings deployed along the line parallel to the conti­ most significant, is rather high compared to that nental slope and with the moorings installed along the observed during Tourbillon (Mercier, 1983): the ratio line normal to the slope. No significant differences KE (bottom)/KE (4000 m) is 1.5 at EDYLOC white appear in the results which could permit any assertion KE (4000 m)/KE (3000 m) was 1.3 during Tourbillon about a polarization by the slope except that it seems (for a greater levet difference). to be small. Energy frequency analysis shows that the intensification is greater for high frequencies. The hypothesis of topo­ graphie Rossby waves to explain this does not hold: firstly, no slope appears on the local topography and secondly, horizontal coherences show that only low frequency oscillations are coherent white those of high Acknowledgements frequency, where topographie Rossby waves are expected, are not coherent (Mercier, pers. comm.). Thus, the explanation for this intensification must be This work would not have been possible without sought elsewhere, and further comparisons with models the technical expertise of G. Auffret, A. Billant, J.-P. will be made. Particularly, a non-linear model taking Gouillou, F. Madelain, R. Perchoc in preparing the account of the topography (Treguier, 1987), without current meter moorings, the CTD and the nephelome­ boundary layer, predicts a KM deep layer intensificat­ ter. The help of the crew of the N/0 Le Noroit, who ion. Thus, the bottom intensification encountered in moored and recovered the moorings under bad weather our experiment would not necessarily be due to a conditions, was particularly appreciated. 1 thank boundary layer effect and there is not enough data to H. Mercier and A. Colin de Verdière who helped me determine whether our deepest levet of measurement is with sorne computer processes. The comments and sug­ or not in the boundary layer. gestions of H. Mercier are gratefully acknowledged.

REFERENCES

Armi L, Millard R. C. Jr, 1976. The bottorn boundary layer in the nepheloid layers in the world ocean, Neth. J. Sea Res., 20, deep ocean, J. Geophys. Res., 81, 4983-4990. 2/3, 167-181. Bendat J., Piersol A., 1971. Random data. Analysis and measurement Mercier H., 1983. Étude de la variabilité d'échelle moyenne dans procedure, Wiley-Intersciences. l'Atlantique Nord-Est à partir des données de l'expérience Tourbil­ Biscaye P. E., Eittreim S. L, 1977. Suspended partic':'late loads and lon, Thèse 3" cycle, Univ. Bretagne Occidentale, Brest, 139 p. transport in the nepheloid layer of the abyssal Atlantic Ocean, Mar. Mercier H., Colin de Verdière A., 1985. Space and Time Scales of Geol., 23, 155-172. Mesoscale Motions in the Eastern North Atlantic, J. Phys. Bowden K. F., 1978. Physical problems of the benthic boundary Oceanogr., lS, 171-183. layer, Geophys. Surv., 3, 3, 255-296. Nyffeler F., Godet C. H., 1986. The structural parameters of the Dickson R. R., McCave 1. N., 1986. Nepheloid layers on the conti­ benthic nepheloid layer in the northeast Atlantic, Deep-Sea Res., 33, nental slope west of Porcupine Bank, Deep-Sea Res., 33, 6, 791-818. 2, 195-207. Dickson R. R., Gould W. J., Muller T. J., Maillard C., 1985. Esti­ Saunders P. M., 1983. Benthic observations on the Madeira Abyssal mates of the mean circulation in the deep ( > 2000 rn) layer of the Plain: currents and dispersion, J. Phys. Oceanogr., 13, 8, 1416-1429. eastern North Atlantic, Progr. Oceanogr., 14, 103-127. Treguier A. M., 1987. Effet des vents fluctuants et de la topographie Eittreim S., Ewlng M., 1972. Suspended particulate matter in the sur la turbulence océanique à moyenne échelle, Thèse Doct. Etat, deep waters of the North American Basin, in: Studies in Physical Univ. Bretagne Occidentale, Brest, 146 p. Oceanography, edited by A. L. Gordon, Gordon and Breach, New Vangriesbeim A., 1985. Hydrologie et circulation, in: Peuplements York, N.Y., 123-167. profonds du Golfe de Gascogne, edited by L. Laubier and C. Monniot, Gurbutt P. A., Dickson R. R., 1986. A vertical eddy diffusivity IFREMER, Brest, 43-70. estimate from near-bottom radon measurements. Interim oceanog­ Vangriesbeim A., 1986. Dynamique et hydrologie de la couche pro­ raphie description of the North-East Atlantic site for the disposai fonde dans l'Atlantique Nord-Est: campagnes EDYLOC, Campa­ of low-level radioactive wastes. Vol. 2, NEA/OCDE, Paris. gnes Océanographiques françaises no 2, IFREMER, 112 p. Harvey J., Glynn S., 1985. Water mass structure and transport in Vangriesheim A., Gouillou J.-P., 1987. Étalonnage du néphelémètre the Tourbillon eddy, Deep-Sea Res., 32, 6, 675-695. grande profondeur de l'IFREMER. Rapport interne IFREMER/ McCave 1. N., 1986. Local and global aspects of the bottom DERO 87-01/EP, Brest.

158